[Lustre-discuss] SLES 11 SP1 Client rpms built but not working

Andreas Dilger adilger at whamcloud.com
Tue May 10 13:48:00 PDT 2011


On May 9, 2011, at 11:38, <peter.chiu at stfc.ac.uk> <peter.chiu at stfc.ac.uk> wrote:
> The rpms lustre-modules, lustre and lustre-tests were then installed smoothly without any complaints.
>  
> But the subsequent “modprobe lustre” will return a “Killed” message, with no lustre module loaded.
>  
> dmesg also reveals  “BUG: unable to handle kernel NULL pointer dereference at 0000000000000008”
>  
> A second modprobe lustre command will then hang, again with no module loaded.
> Subsequently the client is not able to mount the lustre storage.
>  
> Can anyone shed some light as to what has gone wrong here please?
>  
> ./configure --with-linux=/usr/src/linux --with-linux-obj=/usr/src/linux-2.6.32.29-0.3-obj/x86_64/xen

Are you sure that "/usr/src/linux" points to the same source as "/usr/src/linux-2.6.32.29-0.3-obj"?  Is that a symlink?  Normally the source and -obj files have a very similar pathname (i.e. just with "-obj" suffix difference).

> > [  168.647996] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
> > [  168.648066] Pid: 3445, comm: modprobe Tainted: G          N  2.6.32.29-0.3-xen #1
> 0000000000000400
> > [  168.648110] Process modprobe (pid: 3445, threadinfo ffff88007efa4000, task ffff88007e9100c0)
> > [  168.648129] Call Trace:
> > [  168.648138]  [<ffffffff80038588>] try_to_wake_up+0x48/0x420
> > [  168.648143]  [<ffffffff8005b2e8>] up+0x48/0x50
> > [  168.648153]  [<ffffffffa0230d92>] LNetInit+0x92/0xc0 [lnet]
> > [  168.648167]  [<ffffffffa02430ac>] init_lnet+0x4c/0x280 [lnet]
> > [  168.648178]  [<ffffffff80004045>] do_one_initcall+0x35/0x1b0
> > [  168.648184]  [<ffffffff8006d154>] sys_init_module+0xe4/0x270
> > [  168.648189]  [<ffffffff80007458>] system_call_fastpath+0x16/0x1b
> > [  168.648194]  [<00007f3f40bc9f7a>] 0x7f3f40bc9f7a
>  
> I have tried Lustre-1.8.4, but got the same result.
> I have also tried to follow the 1.8 Operations Manual to locate the diagnostic tools, but the link wiki.lustre.org is no longer valid.

This looks like a pretty serious error to oops during module insertion, and I'd suspect the build environment before any particular Lustre code.

Cheers, Andreas
--
Andreas Dilger 
Principal Engineer
Whamcloud, Inc.






More information about the lustre-discuss mailing list